04-25-multiple-regression-3d.knit

Univariate Linear Regressions


	vote share
	(1)	(2)

facial competence	0.660^***
	(0.127)

post 2003		0.088
		(0.054)

Constant	-0.312^***	-0.039
	(0.066)	(0.038)


Observations	119	119
Adjusted R²	0.180	0.014

Note:	p<0.05; p<0.01; p<0.001

Multiple Regression Plane

Hover over a point to see which state that election was held in.


	vote share
	(1)	(2)	(3)

facial competence	0.660^***		0.673^***
	(0.127)		(0.126)

after 2003		0.088	0.101^*
		(0.054)	(0.048)

Constant	-0.312^***	-0.039	-0.369^***
	(0.066)	(0.038)	(0.071)


Observations	119	119	119
Adjusted R²	0.180	0.014	0.203

Note:	p<0.05; p<0.01; p<0.001

Interaction Terms (ie somewhat nonlinear regression surfaces)

This is just for fun, demonstrating what’s called an interaction term. Notice that the green and purple lines are no longer parallel to each other. Interaction terms in a regression allow different slopes for different crosssections of the regression surface. In stats terminology, the marginal effect (slope, kind of) of facial competence can be different after 2003 than it is before 2003 when an interaction term is used in the regression. These are also called heterogeneous effects (different effects for different groups).

Facial Competence & Vote Share

Univariate Linear Regressions

Multiple Regression Plane

Interaction Terms (ie somewhat nonlinear regression surfaces)